Src becomes activated in response to various extracellular stimuli, but the hows and wheres have been somewhat hazy. Thanks to the use of a Src fusion protein, Sandilands et al. have now found that Src becomes activated during transit from the perinuclear region to the plasma membrane. The actin cytoskeleton is required for this, as are RhoB-associated cytoplasmic endosomes.

Once the authors had established that the green fluorescent protein (GFP) tag that they engineered onto Src (to make Src–GFP) didn't interfere with the normal spatial distribution, activation or interactions of Src, they continued with the main aim of their studies to address the connection between catalytic activation and spatial localization. They saw that Src activity increased from the perinuclear region, where it was inactive, to peripheral membrane structures, where it was highly active. Artificially disrupting actin filaments inhibited both the membrane translocation and activity of Src. And when a mutant version of suppressor of cAR (SCAR)/WAVE1 was used to inhibit actin nucleation, Sandilands et al. showed that Src–GFP couldn't translocate to the plasma membrane in response to stimulation by platelet-derived growth factor (PDGF). This hinted that 'outside–in' signals from the PDGF receptor that induce the translocation/activation of Src required signalling to the actin-polymerization machinery.

But the results didn't preclude the idea that endosomes — actin-associated structures with which Src has previously been reported to colocalize — might be involved in this 'outside-in' signalling. Indeed, the authors showed that active Src–GFP colocalized with RhoB, an endosome-associated Rho GTPase, in the perinuclear region and in discrete intracellular structures that surround this region. In the absence of RhoB, Src failed to accumulate at the membrane or be activated in response to cell plating on fibronectin (a main Src-activating extracellular stimulus).

So, could there be a connection between these endosomal structures and the actin-polymerization machinery? Immunostaining showed that Src–GFP and fluorescently tagged forms of RhoB and SCAR/WAVE1 all localized to discrete cytoplasmic structures. Disrupting actin filaments didn't alter RhoB endosomes from cells that lacked Src and the Src-family proteins Fyn and Yes — that is, until Src was re-expressed, in which case RhoB endosomes remained in the perinuclear region. This implies that the presence of Src in RhoB endosomes somehow resulted in them being dependent on the actin cytoskeleton to move. When Src was induced to move to the peripheral membrane at the same time as actin polymerization was re-initiated (both by external means), motile Src- and RhoB-containing structures that resembled endosomal vesicles were associated with newly formed small bundles of actin — actin clouds — between the perinuclear region and the cell membrane. The actin clouds weren't seen when the mutant SCAR/WAVE1 construct was expressed, implying that Src is promoting the actin-nucleation events that are associated with the endosomes.

The authors, therefore, postulate the following model: inactive Src is usually present around the perinuclear region, from where it is 'picked up' and transported to peripheral membranes in RhoB-associated, Rab11-dependent endosomes, probably by inducing actin-nucleation events. Details of exactly how Src induces these actin-nucleation events are awaited.